Does anyone know how i can disable the driver cache or adjust it better?
I have
/ ext 2
everything else ext4
and my storage drive has xfs.
Anytime i need to do a reset, whatever is in the cache is unreadable afterward.
I like to avoid this. The best would be not to have a reset, but sometimes it just happens.
Can i do that?
On 06/11/2011 09:36 AM, JoergJaeger wrote:
>
> Does anyone know how i can disable the driver cache or adjust it better?
what is a “driver cache”, where does it live and what does it do?
> I have
> / ext 2
> everything else ext4
> and my storage drive has xfs.
sorry, i know nothing about xfs…
> Anytime i need to do a reset
what is a “reset” any why would anyone need to do that?
oh, you mean a brutal power switch shutdown???
no no
> whatever is in the cache is unreadable
> afterward.
> I like to avoid this.
remove the need to ‘reset’
> The best would be not to have a reset, but
> sometimes it just happens.
should not! what is the recurring problem which does??
have you heard of REISUB?
google it:
http://www.google.com/search?q=site%3Aforums.opensuse.org+REISUB
–
dd CAVEAT: http://is.gd/bpoMD
via NNTP openSUSE 11.4 [2.6.37.6-0.5] + KDE 4.6.0 + Thunderbird 3.1.10
Acer Aspire One D255, 1.66 GHz Atom, 1 GB RAM, Intel Pineview graphics
- When your gecko is broken you have a reptile dysfunction! *
I had the same sort of questions DenverD has. But first I thought: “you talk about a subject that is not within my knowledge base”. But in fact I wanto be illuminated also. What do you reset (and how), which driver with what cash and so on.
I realy do not understand what you are doing/seeing/clicking/typing.
Only thing I could advice is: convert your / file system to at least ext3, better ext4. But I guess that has nothing to do with your problem.
I am not sure how to describe it in an other way.
From my understanding, not all datas are written right away onto the harddrive, so some data are held back in a cache.
If all of a sudden your computer freezes or resets, these datas in the cache are not written onto the harddrive and are lost, resulting in a defective or incomplete file.
Isn’t it with a usb stick the same. You don’t suppose to remove it right away and rather eject it so that all datas are really written on it.
So i used the term drive cache, but it might be something else. Not sure.
It might be a buffer or some form of data held in memory.
That is I/O caching. without it your Linux would run like a snail.
Linux systems are known for their stability and seldom “freeze”. A reboot might be possible, but a “reset” (whatever that may be)? Power failures are the worst thing that can happen. When your area is prone to those, you should think abbout an UPS. When power fails it will go to battery and warn your system which will then do a proper shutdown on the battery power. And when you have children that are used to pulling the power cord out of the wall, you should handle that internaly.
I need to understand how linux does this. It did not sink in so far, but i can see what you mean.
My thinking was, of course, that i can write it directly onto the harddrive to avoid it. But that is then counter intuitive i assume and defeating the purpose.
Its not happening all the time, just somtimes if for example a program crashes and takes the system with it. But it is not an overall issue.
If a program crashing causes the kernel to crash then you need to remedy that.
On 06/11/2011 11:06 PM, JoergJaeger wrote:
>
> just somtimes if for example a program crashes and takes the system
> with it.
that should not happen…yes, i know it does sometime…but, it should
be SO very seldom that when it happens your very next action should be
to find out why…what was the problem…
anyway, what do you mean “takes the system with it”, do you mean it
crashes the desktop? if so, Linux may still be working fine, just the
desktop needs to be restarted
well, one thing i’ve learned is that many of the times i think my system
is ‘frozen’ (if that is what you mean by crash and take system along)
all i need to do is walk away for the machine and give it some time
(WITHOUT me clicking, tapping, pushing, etc etc etc…just leave it
alone)…maybe 15 minutes…
if it is still ‘frozen’ then try ctrl+alt+F1, log in and start top to
see who the hog is…and, kill that one app…
if you can’t ctrl+alt+F1 then, do ctrl+alt+backspace+backspace (to kill
and restart desktop)…
if you can’t do that then do SysRq+R+E+I+S+U+B to softly shut down the
system WHILE flushing the cache to disks…
all of those give you a chance to avoid loosing data
–
dd CAVEAT: http://is.gd/bpoMD
via NNTP openSUSE 11.4 [2.6.37.6-0.5] + KDE 4.6.0 + Thunderbird 3.1.10
Acer Aspire One D255, 1.66 GHz Atom, 1 GB RAM, Intel Pineview graphics
- When your gecko is broken you have a reptile dysfunction! *
@Denver Excellent advice. I will note that in my own linux book that i keep. 
There are some things i am just not aware off and something like this, might already help in avoiding disasters.
Thanks for that.
One time it was pan which has a limitation of how much you can shove in the memory. The other was Gnome 3 which froze my desktop. It was just blank and i wasn’t able to do anything. Well, the fallback to the terminal i wasn’t aware off. So i will try that next time.
On most occasions its nothing of importance thats gotten lost. Its just annoying.
As regarding pan. I just avoid to load to much. That solved already one of the culprit.
On 06/12/2011 01:36 AM, JoergJaeger wrote:
>
> One time it was pan which has a limitation of how much you can shove in
> the memory.
hmmmm…i used pan for years and don’t recall it crashing or having a
memory problem…
now, i admit i’ve not used it in the last six or seven years so it might
have been ‘innovated’ and ‘improved’ to the point it is now broken, but
i kinda doubt it…
[well, as i recall i quit using it when the developer started messing
with the UI too much, for me…i didn’t want to have to relearn
what-is-where and how-to-do every release…]
if it were me i’d join a pan mail list <http://pan.rebelbase.com/>, or
look at their bugs, or something to learn what is wrong and how to fix
it…or move to a nntp reader which works…
bottom line: do not use applications which crash the kernel, instead log
a bug.
btw: thunderbird (with all its bloat and warts) has done all the nntp
tasks i’ve needed of it for several years…it does sometimes just fold
up and quit–but, it seems to always manage to save stuff and restart
smoothly without freezing the DE or crashing the kernel. ymmv
–
dd CAVEAT: http://is.gd/bpoMD
via NNTP openSUSE 11.4 [2.6.37.6-0.5] + KDE 4.6.0 + Thunderbird 3.1.10
Acer Aspire One D255, 1.66 GHz Atom, 1 GB RAM, Intel Pineview graphics
- When your gecko is broken you have a reptile dysfunction! *
Pan quite works a lot if the group has over a certain amount of news headers.
Anything beyond the millions makes it work hard. Problem is, i am inpatient. The workload can take over the whole computer so that nothing works anymore or responds.
I gave up on that and just don’t do it anymore.
My guess is, that it is busy swapping around.
On 06/12/2011 11:36 PM, JoergJaeger wrote:
>
> Anything beyond the millions makes it work hard.
oh! hmmmm…ok, you seem to be using nntp in ways i never have…
are all those millions of messages still kept on your usenet providers
server? amazing, mine flushes out everything after a few months of
retention, and so me holding the header is not useful for me as i can’t
use it to once again collect the header’s body/message…
–
dd
http://is.gd/bpoMD
In any case, switching off the power of a computer (if that is what yi omean with the word “reset”, I call it “murder”) is not a remedy against an application monopolizing the system. You kill the culprit, not the system. When you overload you car with so much load that it does not move uphill any more, you do not drive it onto the canyon, but you unload some of it to get it up the hill later.
I do not know if this is a sort of leftover of the times you had Windows on that system (when you ever experienced Windows) but it looks very much like I heard of those “users”, when it does not respond for a minute, switch it off.
But have you ever heard about an IBM mainframe switched off because one user thought his application took more time then he scheduled for it? And Unix/Linux is much more like a stable manframe the a Windows playground for children. Please handle you computer with the care it asks for.